Soft Sweeps Are the Dominant Mode of Adaptation in the Human Genome

Abstract

The degree to which adaptation in recent human evolution shapes genetic variation remains controversial. This is in part due to the limited evidence in humans for classic "hard selective sweeps", wherein a novel beneficial mutation rapidly sweeps through a population to fixation. However, positive selection may often proceed via "soft sweeps" acting on mutations already present within a population. Here, we examine recent positive selection across six human populations using a powerful machine learning approach that is sensitive to both hard and soft sweeps. We found evidence that soft sweeps are widespread and account for the vast majority of recent human adaptation. Surprisingly, our results also suggest that linked positive selection affects patterns of variation across much of the genome, and may increase the frequencies of deleterious mutations. Our results also reveal insights into the role of sexual selection, cancer risk, and central nervous system development in recent human evolution.

Keywords: adaptation; machine learning; population genomics; selective sweeps; soft sweeps.

Fig. 1.

Enrichment of various annotation features in regions classified as sweeps or linked to sweeps relative. The fold enrichment is the ratio of the number of base pairs in the intersection between windows assigned to a given class and an annotation feature divided by the mean of this intersection across the permuted data sets (see Materials and Methods). This was calculated separately for each population. (A) Enrichment of elements in windows classified as hard sweeps. (B) Same as A, but for soft sweeps. (C) Enrichment of elements in windows classified as affected by linked hard sweeps. (D) Linked soft sweeps.

Fig. 2.

The number of windows assigned to each class by S/HIC in each population.

Fig. 3.

Enrichment of pairs of interacting genes each falling within a window classified as a sweep. The fold enrichment is the ratio of the number of pairs of interacting genes overlapping a window classified as a sweep of a given type divided by the mean of this number across the permuted data sets (see Materials and Methods). This was calculated separately for each population. When no pairs of interacting sweep genes were observed in our true data set or a population, no bar was drawn. (A) Enrichment of pairs of genes encoding protein products that physically interact with each other (data from BioGRID) and both overlap hard sweep windows. (B) Same as A, but for soft sweeps. (C) Enrichment of pairs of genes, one of which is encodes a transcription factor that affects expression of the other (data from ORegAnno), where both overlap hard sweep windows. (D) Same as D, but for soft sweeps. (E) Enrichment of pairs of genes for which a genetic interaction has been observed (data from BioGRID) and both overlap hard sweep windows. (F) Same as E, but for soft sweeps.

Fig. 4.

Hard selective sweep near several SPATA31 spermatogenesis-associated genes. The S/HIC classification tracks show the raw classifier output for each population (red = hard sweep, blue = soft sweep, light red = hard-linked, light blue = soft-linked, black = neutral). We also show the values of various population genetic summary and test statistics (π, Tajima’s D, Kelly’s Z_nS, and the SweepFinder composite likelihood ratio, or CLR). To avoid clutter, we only show statistics from CEU.

Fig. 5.

Soft selective sweeps near CADM1. The same tracks are shown as in figure 4.